At the Ref Desk (7/11/20): After 6 hrs of chat-reference, I got out of the house and hiked north on Broadway all the way from Diversey to Howard. Nice evening. (Took the bus home.) [more...]

Library Webpage Design and the Notion of Conversion Rates

Submitted by Leo Klein on Sun, 8/19/07 (1:45pm)

We're all familiar with usability. Basically it's a way of assessing the success rate of any one task. Successfully finding a book or journal article starting from the home page would be a typical task to measure.

I'm wondering if it might also be helpful to think of task completion in the way marketers do, namely as "conversion rates".


When librarians use the work "marketing", usually they mean 'getting the word out'. Marketers go one step farther: marketing for them means actually selling a product.

This notion of a complete transaction can be useful.

We all have user populations. When these users come to our site, they represent potential "sales" of our products and services.

This is where conversion rates come in.

(more after the jump...)


In usability, we're measuring success based on a limited sampling ("five out of ten users successfully completed the task"). With conversion rates we're measuring actual use of our products and services based on a percentage of total web traffic.

A lot of people visit our site. Some of them use our stuff. The goal is to increase the percentage of those using our stuff through incremental improvements to the design of our pages.

You can see the dynamic at work in a study by (MEC) called "Landing Page Confusion".

In it, they look at improving product pages (i.e. "landing pages") though changes in font, color and layout. If the changes have a positive impact on the number of widgets, subscriptions, or whatever that are sold, then it's considered a success.


Naturally in the library context, such things are harder to measure. Our systems aren't integrated and this makes following our users from start to finish a bit of a challenge.

Yet we can begin to develop an overall sense by looking at the indices that we do have, namely from our own site and from vendors.

By mixing and matching this data and by slightly tweaking the models, we can develop an idea of whether our sites are having a positive effect, not simply in themselves but end-to-end though the entire research process.